Low-rank Low Precision Training of Gaussian Processes (Q7424482)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

No description defined
Language Label Description Also known as
default for all languages
No label defined
    English
    Low-rank Low Precision Training of Gaussian Processes
    No description defined

      Statements

      Low-rank Low Precision Training of Gaussian Processes (English)
      0 references
      9 November 2026
      0 references
      math.NA
      0 references
      cs.LG
      0 references
      Moritz Poguntke
      0 references
      Eda Oktay
      0 references
      Yuanzhe Xi
      0 references
      Erin Carson
      0 references
      Martin Stoll
      0 references
      We develop a scalable low-rank GP training method using AnchorNet landmark selection with FITC correction that reaches reference predictive accuracy while substantially reducing training time for large-scale regression datasets. The required approximation rank varies across problems, but single precision is limiting for ill-conditioned cases and lower-precision variants provide modest additional speedups at the cost of reliability.
      0 references
      Low-Rank
      0 references
      Gaussian Processes
      0 references
      Scalable Learning
      0 references
      Approximate Inference
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references